A STUDY OF THE ATOMIC WEIGHT OF INDIUM.1

compared with a standard solution of titanium in the colorimeter. The following results were obtained: Volume of solution. c.c.. Ti02 present gms. HF1...
1 downloads 0 Views 696KB Size
A S T U D Y O F THE A T O M I C W E I G H T O F I N D I U M

485

potassium fluoride were compared with a standard solution of titanium in the colorimeter. The following results were obtained : Volume

of solution C.C.

Ti02 preseiit gms.

H F1

present gms.

I

250

0.010

O.Oo0

2

250

0.010

3 4

2 50

0.010

250

0.010

0.00039 0.00194 0.-39

Colorimeter reading obtained by comparing solution with ( I ) a s standard.

100.0 93.3 80.2

67.6

Similarly, the change in color caused by adding different amounts of phosphoric acid to a titanium solution of the same concentration was determined. T h e phosphoric acid used was prepared by Kahlbaum. The titanium solution contained 5 per cent. sulphuric acid and 2 cc. of 3 per cent. hydrogen peroxide in every 250 cc. Volume

of solution C.C.

TiOn present gms.

P2Oj present gms.

Colnrimeter reading obtained by comparing solutionwith ( I ) a s standard.

I

250

0.010

0.00

100.

2

250

0.010

0.13

3 4 5 6 7

250

0.010

0.26

90. 83.

250

0.010

0.52

250 250

0.010 0.010

0.78 I .04

250

0.010

1.30

74. 69. 66. 64.

That the change in color was not due to an impurity in the phosphoric acid used was proved by the fact that on repeating the experiment with solution (3) and using in 25c cc. of solution 0.638 g. of pure K,HPO, (containing 0.26 g. P,O,) instead of phosphoric acid a reading of 82.5 was obtained, which agrees with the value obtained when phosphoric acid was used. Moreover, an experiment in which phosphoric acid obtained from another source was used gave results agreeing with the above. T h e influence of the phosphoric acid on the titanium color is of importance in the colorimetric determination of titanium in substances containing any considerable amount of phosphates. The results obtained will be low unless enough phosphoric acid is added to the standard to compensate the effect of that in the sample.

A STUDY OF THE ATOMIC WEIGHT OF INDIUM.' FRANK C U R R Y M A T H E R S .

Received January

25,

1907.

Historical. T h e first determinations of the atomic weight of indium were made by Reich and Richter2, who obtained values that varied from I I O to 112. 1 NoTE.-The following investigation by Mr. Mathers upon t h e Atomic Weight of Indium is a continuation of the work t h a t was first taken up in t h e Cornel1 LaboL. M. DENNIS. ratory by Mr. W. C. Geer. Reich and Richter, J. pr. Chem., 92,480 (1864).

486

F R A N K C U R R Y MATHERS

They oxidized metallic indium \\it11 nitric acid and determined the ratio of indium to indium oxide. T h e same authors determined the ratio of indium sulphide to indium oxide a i d barium sulphate. They macle only a few experiments and the methods that they follo\vecl have since heeii shown to be very inaccurate. liichartls has demonstrated that it is impossible to free precipitated bariuin sulphate irom occluded salts' a i d that metallic oxides that are prepared b!. the ignition of tlie nitrates always retain some occluded gases'. T o frirther complicate the ivurk, indium oxide is appreciabl!. volatile a t I W O ' ~ , and is also hygroscopic. From the ratio of indium to gold, l\.inkler4 obtainetl the value, I 14. This method should have given correct results, if tlie material had been pure. -A siiiiilar ratio, copper to silver, was determined 1,- Richards~and esccllent results \Yere obtained. i\.inkler made only t t v o determinations ; coiisequently, tlie result cannot be gi\-en great \veight. The same author" also obtained the values 107.8and I 13.8 from the ratio o i indium to intlium osiclc. 13iinsen' incidentally made t\vo determinations of the atomic \\eight of indium in connection Jvith his ivork iipoii the specific heats of the elements and upon calorimeters. H e ascertained the ratio of indium to indium oxide and obtained the value 113.8. T h e only recent umrk upon tlie atomic weight of indiuiii is that performed by Tliiclq. Eight determinations hy him of the ratio of indium to indium oxide gave valucs tliat vary from I 1 3 4 to I 15.6. Since this ratio is subject to the inaccuracies inentioiied above, the results obtained are of slight importance. Froni the ratio of indium to indium trichloride and of indium to indium triiodide, Thiel obtained values that approximate I I j.4. It is quite probable that this figure is too high owing to errors caused bp loss from volatilization during drying. l ~ ythe presence of basic compountls in the trichloride or triiodidc, or by the presence of moisture or foreign salts in the spongy metallic indium that had been only lightly pressed. Thiel also determined the ratios of indiuiii trichloride to silver chloride and of indium tribroiiiide io silver bromicle. H i s procedure \vas as fol1on.s :--To i?-eiglied portions of indium trichloride or tribromitle there was added a slight escc'ss of a standard solution of silver nitrate. -4fter filtl-ation, the amount of this cxcess \\.as determined by precipitating the silver in the filtrate with hydrochloric acid and weighing this precipitate of silver chloride. The weight of tlie chlorine that was in combination ivith tlie indiiun was ascertained by subtracting the u.eight of

' Ricllarcls,

Z. artorg. Cheni., 8, 413 ( 1 8 9 j ) . Richards, Proc. Amer. hcad. Arts and Sci., 26, 281 ( ~ S g o - ~ S g r ) . :'Meyer, %. anorg. Chern., 47, ZSI ( 1 9 o j ) . ' Winkler, J. pr. Chetn., 102, 2;3 ( is67 1. "Richards, 4 m . Ch. J , , IO, 182 (rSSS). "J. pr. Chem. 94, I (1665). ' Pogg. Ann., 141,2 S (1870). Z.anorg. Chem., 40, zSo (r934).

A STUDY O F THE ATOMIC WEIGHT O F I N D I U M

487

the silver of the silver chloride from the total silver that had been added as silver nitrate. No correction was made for the solubility of silver chloride. The average of four results obtained from the ratio of indium trichloride to silver chloride is 115.05. Three results from the ratio of indium tribromide to silver bromide, give an average of 114.81. Braunerl recalculated Thiel’s results obtained from the ratio of indium trichloride to silver chloride, and, by using a “somewhat arbitrary” correction for the solubility of silver chloride, obtained the value I 14.9. The International Committee on Atomic Weights has adopted the value 115. Below is a tabulated list2 of the various determinations. Name

Year

Reich and Richter

1864

.........

1865 rS67

Winkler

.......... ............

Bunsen Thiel

1870 I904

Ratio

nIn :In,O, I n & : In,Ol :3BaSO, nIn :In,O, I n :Au zIn :In,O, 2In :In,O, nIn : In,O, I n :InCI, I n :InI, InCl, : 3AgCl InBr, : 3AgCI : 3AgRr

Atomic Weight In 111.2 112.

107.8 114 03 113.58 113 84 I 13.4-115.6 115 4 115.3 115.01

114.83

T h e standards used in the table are : silver, 107.92; chlorine, 35.47; and bromine, 79.955. Experimental. The wide variations in the earlier results upon the atomic weight of indium led Dennis and Geer to take up the study of the subject in this laboratory in 1903. i\-hile that investigation was still in hand Thie13 published a preliminary note in which he announced the provisional atomic weight of I 15.08. This called forth a preliminary note4 by Dennis and Geer in which they describe new methods for the purification of indium and the preparation of the metal. Thie15 subsequently published full results upon the ratios, InCl, to 3AgC1 and InBr, to gAgUr. By the first ratio the atomic weight of indium was 115.05; by the second, 114.81. Although his work was undoubtedly niore accurate than any that had preceded it, the averages of the results obtained by the two ratios differed by 0.24. Consequently the study of the subject was continued in this laboratory with the results detailed below. Balance aizd W’eiglzts. The balance used in this work was manufactured by A. Ruprecht. One niilligrani shifted the zero point ten divisions. This sensitiveness could be increased to such an extent that 0.1 milligram would shift the zero point eight spaces, but this degree of sensitiveness was not Rrauner, Abegg’s Handbuch, 3, Part I. Ibid. ’Ber. 37, 175 (1904). Deniiis and Geer, this Journal, 26, 437 (1904). b Z . anorg. Chem., 40, 280 (1904).

i

488

FRAXK C U R R Y MATHERS

required for the work. All weighings were made by the method of substitution and were reduced to vacuum. The weights were standardized by comparison with quartz weights, which, within the past year, had been standardized by the Bureau of Standards at Washington. The balance was kept in a “constant temperature room” where the total variation in temperature during the entire period of work amounted to only a few degrees. T h e table that supported the balance rested upon the solid concrete floor. Vessels containing concentrated sulphuric acid were used to keep the room dry. Calcium chloride was used inside of the balance case. Pwificatio?z of Materials. Iizdinni. Conimercial metallic indium was dissolved in less than the required amount of hydrochloric acid, and the solution was filtered from the residue of indium and such metals as copper, mercury, silver, etc. A n excess of hydrochloric acid was added and the solution was treated with hydrogen sulphide. The insoluble sulpliides were filtered out and the hydrogen sulphide was removed by boiling. Ammonium chloride and ammonium hydroxide were added and the solution was warmed. Indium hydroxide was precipitated, together with the hydroxides of iron, aluminum and probably a small amount of zinc, T h e yreater part of the zinc remains in solution. After washing with water, :he precipitate was dissolved in hydrochloric acid and reprecipitated with ammoniuni hydroxide and ammonium chloride and again washed with water. This solution in acid and reprecipitation with ammonium hydroxide was repeated four times. These steps should remove zinc. For the removal of the greater part of the iron that was still present, the precipitate was dissolved in hydrochloric acid and the solution was evaporated almost to a syrup. This was transferred to a separatory funnel, an excess of potassium sulphocyanate was added and the red ferric sulphocyanate was removed by extraction with successive j o cc. portions of ether‘. T h e ndium was then precipitated as the hydroxide with ammonium hydroxide and this was washed with water. The indium hydroxide n a s next dissolved in hydrochloric acid and the solution was evaporated until almost all of the free acid was removed. This solution was placed in a beaker and the metallic indium was deposited by electrolysis with a current of 0.j ampere. T h e platinum atiocle was placed in a porou5 cup and the platinum wire cathode extended only 0.25 crn. into the liquid. T h e beaker &vaskept covered with a bell jar and the chlorine was removed by suction :hrough a glass tube, which reached under the cover glass of the beaker. rhis very high current density at the cathode lasts for only a very short time because the metallic indium soon grows down into the electrolyte and increases the cathode surface. If a larger platinum cathode n-ere used, nluch of the indium deposit could not be mechanically removed, since it adheres Under the conditions above - - very tightly to the platinum . - . ._surface. ___--described the indium is deposited in a crystalline form and is soft and I

Dennis and Geer, this Journal, 26, 437 (Ical\.

A STUDY OF THE ATOMIC WEIGHT O F I N D I U M

489

malleable. The compactness of the metal bears a direct ratio to the amount of acid present. T h e metal was washed with water and compressed, if it was very crystalline and soft. After washing it with alcohol and drying a t I ~ o "the , metal was fused in a charcoal crucible in an atmosphere of hydrogen. T h e temperature required was not high, as indium melts at 155'. T h e metal thus purified should be free from aluminum, chromium, manganese and such metals as are not electrolytically precipitated from an aqueous solution. Since indium is electro-negative to zinc and cadmium, these metals also should not be deposited with the indium during electrolysis. Nevertheless, the indium was found to contain traces of iron and, consequently, it was further purified by fractional electrolysis from a solution of indium sulphate acidified with sulphuric acid. The electrolysis was carried on in a large test tube through the bottom of which was fused a platinum wire. The metallic indium to be purified was placed in the tube directly in contact with the platinum wire which was made the anode. The indium was covered to a depth of about four centimeters with dilute sulphuric acid ( I : 4 by volume) and a platinum wire cathode was immersed in the liquid to a depth of about three millimeters. The indium is deposited on the cathode in the form above described. A considerable quantity of fine anode slime is produced, but this remains at the anode and does not contaminate the indium that separates upon the cathode. T h e fractionation vias carried on twice to effect the complete removal of the iron. The metallic indium was removed from the cathode, was washed with water and with alcohol and was dried. It was then fused in a charcoal crucible in a current of hydrogen to remove occluded salts and water. I t was repeatedly observed that the metallic indium obtained by electrolysis gave off upon fusion appreciable amounts of water even when it had previously been dried for some time in an air-bath. T h e fused metal was cleaned with dilute hydrochloric acid and was rubbed with filter paper. T o ascertain whether the metal was free from iron, a portion of it was converted into anhydrous indium bromide. 3.2323 grams of this salt was dissolved in 15 cc. of water and placed in a Nessler tube of 25 cc. capacity. Ten cubic centimeters of normal ammonium sulphocyanate was then added. T h e solution showed no appreciable color. L p o n adding to this solution o.000~gram of ferric chloride, a distinct red color appeared, Consequently, if the indium bromide contained iron, the amount of the latter was less than 0.006 per cent. This metallic indium thus purified was used in the preparation of the different compounds employed in the determination of the atomic weight of the element. Byomiize. "Pure" bromine, purchased of Kahlbauni, was shaken with about one-fourth of its weight of calcium bromide and a few grams of zinc oxide' and was allowed to stand in contact with these materials for 'Stas, Mem. de l'academie royal des sciences, des lettres, et des beaux-arts de Relgique, 43, Note 3 (1882).

490

F K A S K CUKRI' MATHEXS

several days. By this treatment iodine is converted to zinc iodate' and chlorine to calcium chloride. The bromine was distilled off and the distillate \\-as then subjected to iractional distillation. The middle fractions x e r e emplo!.ed. C'h/miuc. "Pure," concentrated hydrochloric acid \vas diluted with ar? equal volume of water, was then heated to about 8;" ancl \\-as treated \\it11 successive small portions of potassium permanganate. The liquid \vas then hezted to boiling and an)- broiiiine or iodine set free by the potassium permanganate distilled oj'er in the first fractions. The middle fraction was collected separately and was then again subjected to fractional clistillation. The middle fraction of this second distillate \vas reserved for use. The distillation \vas carried on in an apparatus macle entirely of glass. Chlorine \\:as generated by heating this acid n-ith manganese dioside that had previously been treated with hot dilute sulphuric acid and had been washed free froin tliat acid. Silvw. Sietallic silver \vas dissolved in nitric acid and \\!as precipitated froin a very dilute solution \\.it11 hydrochloric acid. The silver chloride was n-asheti thoroughly \vith iixter, then with strong 111-drochloric acid. and again with xvater. It \vas nest treated with strong nitric acid and was freed from this acid h!. further washing with \\.ater. This precipitated silver chloride \\-as then transferred to a platinum dish and \vas covered to a depth of about 2 cm. with very dilute sulphuric acid. This was then electrolyzed, n.ith the dish as cathode and a platiiiuiii disc as anode. tlie passage of the current being maintained until all of tlie silver chloride had been reduced. The resultant spongy silver \vas fuscd i i i a charcoal crucihle Xvitli the bhst 13111p. It ]vas iicst cleaned \\-it11 sand and nitric acid. The silver \:-as then iiirther purified by placing it i i i ;I porous cup, covering it xvith a solution o i silver nitrate, and inimersiiiji the cup in a beaker coiitaining more of the solution of silver nitrate. This was tlieii subjected to electrolysis, the sil\.er to be purified serving as anode, and a piece of platinum foil that \vas inserted in the solution in the lieaker serving as cathode. Connection between the metallic silver and the circuit was made by means of a platinum rod. Il'ith a currcnt of about one ampere per IOO sq. cin. of cathode surface the silver is precipitated in loose crystals tliat can easilj- be separated from the cathodc. Adhering silver salts were removed by washing the crystalline silver with dilute animonium hydroxide and then wit11 water. IVhcn ivantecl for use about one-forty-seventh more than the amount of silver required for precipitation was weighed out ancl was fiised with the blast lamp in a charcoal crticible. The silver button \vas cleaned with sand and \villi dilute nitric acid arid the weight ]vas reduced to about one-fiftietli escess by means of a file. 'Richards and Alerigold, %. allorg. Chem., 31,

261 (1902).

A STUDY OF

THE A T O M I C W E I G H T O F INDIUM

49'

Sitric Acid. The purest nitric acid of coninierce was distilled from a flask of Jena glass against the side of a platinuni dish which contained ice water and which acted as a condenser1. The condensed nitric acid was collected and preserved for use in a platinum dish. Carbon Dzoxide. Carbon dioxide was generated in a Kipp apparatus from marble and nitric acid. The gas was passed through a solution of sodium bicarbonate to remove any acid that might mechanically be carried along with the carbon dioxide. -4ny oxygen in the carbon dioxide was removed by passing the gas over red hot copper shot contained in a porcelain tube. Dzstilled Water. T h e distilled water of the laboratory was redistilled after the addition to it of a small amount of potassium permanganate to destroy organic matter. The middle portion of this distillate was again redistilled and the middle fraction of this last distillate was used. Glass stoppered distilling flasks were employed, and the condenser used in the last distillation was of block tin. Drying Apparatus. The drying train for each gas consisted of two gas wash bottles filled with concentrated sulphuric acid, a drying tower containing pumice stone saturated with sulphuric acid, and a tttbe containing phosphorus pentoxide. T h e pumice stone, before using, was digested with dilute sulphuric acid, washed with water and dried. The drying efficiency of the sulphuric acid was shown by the fact that the phosphorus pentoxide would remain perfectly dry even after a current of gas had been passed over it for from two to three days. M E T H O D OF ANALYSIS.

The solution to be precipitated was placed in a 1.1 liter Jena glass flask, 4 to 5 cc. of nitric acid was added, and the solution was diluted with distilled water to about 400 cc. About 1/50 excess of an approximately I per cent. silver nitrate solution was added drop-wise Ivith constant shaking. This silver nitrate solution was made by dissolving the silver in nitric acid and evaporating this solution until the excess of nitric acid was removed. The flasks were allowed to stand for from twenty-four to fortyeight hours until the mother liquor became clear. The precipitate \vas filtered upon a Gooch crucible and was washed by decantation until the filtrate was no longer acid. It was then transferred to the crucible by a stream of water from the wash bottle. That portion of the silver precipitate which adhered to the sides of the flask was reinoved without difficulty by vigorously shaking it with small quantities of pure water. From 1200 to 2000 cc. of wash water was used. All of the above steps were carried out in a photographic dark room to prevent the action of light upon the silver precipitate. The filtrate and wash water were passed through a filter paper to catch shreds of asbestos or any of the precipi'Morse and Burton, Ani. Ch. J.,

IO, 31I

(1888).

49 2

F R A N K CURRY 3IATHERS

tate. In the case of silver chloride, the wash water rvas evaporated a d the silver chloride that was in solution was collected upon a filter paper. After ignition of the paper and the precipitate, the silver was dissolved in nitric acid and was titrated with a standard solution of potassium sulphocyanate. T h e solubility of the silver bromide is so small that it was neglected. T h e silver chloride or silver bromide was dried at 13j0, for few hours and then at 150' for from fifteen to twenty hours, and was weighed. It was then transferred to a porcelain crucible and was fused. Any loss of weight on fusion x a s subtracted from the original weight o i crucible and precipitate. THG RATIO O F I K D I U M TO I N D I U M SULPHATE.

Metallic indium \+as placed in a platinum dish, dissolved in redistilled sulphuric acid, and concentrated on a water bath until crystals appeared. T h e platinum dish u a s then placed in a porcelain dish which was covered by an inverted porcelain dish as a protection from dust. thermometer was inserted between the dishes in such a position that its bulb was in contact with the platinum dish. The heating was carried on upon a hot plate. A temperature of 400' was maintained for forty-two hours, but even after that length of heating it was found that the weight was still decreasing. The temperature was then raised to 4 j o " for fifteen hours, whereupon the weight apparently became constant. The indium sulphate is, however, so very hygroscopic that accurate weighing is very difficult of accomplishn~ent. The results obtained have no real value, but are as follows : (.ranis of Iridium

1.3045 1.0842

C: rani5 of I n2( SO4)

Atornic

T$

eight Iridiiiin

2

96i5

113 o j

2

1659

I 13.08

The incomplete removal of sulphuric acid would cause the results to be too low. If heated with a free flame to a dull red heat, some of the indium sulphate is decomposed with the formation of indium oxide as shown by the yellow color. This oxide volatilizes slightly on continual heating. T H E R.ITIO O F I K D I V M DICIILORIDC TO SILVER CHLORIDE.

Pure metallic indium was placed in one end of a Jena glass tube that was about go cm. long. This tube had previously been constricted so as to form four compartments. After displacing the air with hydrochloric acid gas, the tube was heated with a Runsen flame. T h e hydrochloric acid began to attack the metal at a temperature just below the melting point of indium (155'). The temperature was maintained at such a point that the metal and the chlorides that were formed remained liquid. As long as metallic indium was present, the product was dark brown, almost black, indium monochloride (InC1). Further action of the hydrochloric acid changed this into the light yellow, nearly transparent, indium di-

A STUDY O F T H E ATOMIC WEIGHT O F I N D I U M

493

chloride (InCl,). This indium dichloride was distilled into the second compartment in a current of hydrochloric acid gas. After displacing all of the hydrochloric acid gas with carbon dioxide, the indium dichloride was distilled in a current of the latter gas into the last two compartments. The tube was then cooled and the carbon dioxide was displaced with dry air. T h e two samples were sealed off with a blast lamp, and the tubes were dried and weighed. T h e ends of the tubes, which had been scratched with a file before weighing, were then broken off, the tubes were placed in beakers, and the indium dichloride was dissolved in dilute nitric acid. Indium dichloride reacts with water according to the following equation : 31nC12( H,O) = 2InC1, In. This explains the use of nitric acid in bringing the salt into solution. Even under this treatment, however, the resultant solution was never perfectly clear. This is probably due to the action, even at a low red heat, of indium dichloride upon the glass of the tube. No quantitative determinations were made with solutions that showed this turbidity. T o prevent the corrosion of the glass, the later portions of the indium dichloride were not distilled from one compartment of the tube into another, but were simply sealed off in that portion o€ the tube where they were formed after the hydrochloric acid gas had been completely displaced by carbon dioxide. All of these samples gave clear solutions when the substance was treated with dilute nitric acid. Upon precipitating the chlorine by silver nitrate in the manner previously described, varying results for the atomic weight of indium were obtained. Some of the resulting values are the following :

+

+

Grams of Indium.

Grams of AgC1.

Atomic weight of Indium.

343.54 3.8492 4.6482

5.54176 6.03475 7.2359

112.03 111.99 I 13.2

Low values would be caused by the incomplete removal of the hydrochloric acid gas or by the formation of some indium trichloride. The ratio was abandoned because of the inaccuracies inhering in the method. THE RATIO OF I N D I U M TRICHI,ORIDE TO SILVER CHLORIDE.

Pure metallic indium was heated in a current of chlorine gas in a constricted tube such as described above. Indium monochloride is first formed; this slowly changes to indium dichloride. T h e latter halide is then gradually converted into indium trichloride by the continued action of the chlorine. Indium trichloride is a white substance which sublimes easily at a temperature just below a red heat. The indium trichloride was sublimed twice in chlorine and twice in carbon dioxide. The tube was then filled with dry air and the samples were sealed off. T h e indium trichloride thus prepared gave a perfectly clear solution only when some

494

F R A N K CURRY MATHERS

nitric acid ~ v n sadded to the \ v a t u in \vhicli tlie substance \vas dissolved. Tliis is probahlp to be explained l q tlie fact that not all of the indium clichloride \;.as con\-erted into the trichloride and that, as a coiiscqueiice, srmc metallic indium i v a s set irec \\.lien the sabstancc ~ v z i streated \\-it11 \\xter. It is also possible, ho\\-cvcr, that the change into indiuiii trichloritle u.as coiiiplete, Lnit tliat this coiupound \vas dissociated to some extent iiito int1iu:ii dichloride and cliloriiie'. I t is prcsuinalile that this dissocia tioii o i tlic iiitlium trichloride ivould take place \\-lien :he substance is licutecl iii die ciirreiit of pure carboii dioxide because of the ready removal of the clilorine from the tube and because crystals of the siibstance filling a s tlii.!. t l i t l the entire cross section of the tube v;ould act to some extent ;is a porous tlia~1irag.m. To prek-eiit such dissociation, later snblimations ivere carried oii in a current o f carboii dioxide that contained some chlorine gas. .ifter the sublimation had been effected, the temperature \vas lo\veretl heloiv the sublimation point, antl pure carboil dioxide containing no chloriiic gas \vas then p;isse(l throiigh the tube. This \vas continued for M i aii hour and the tube \,vas then alloived to cool, the passage of the carljon dioxide through tlie tube being continued through tlie cooling aiitl for from eighteen to tiveiitj- hours after the tube hat1 reached rooni tcniperature. The carlmii clioxitle \vas then tlisplacctl hy air and the tube sealed oiF at its constrictions. Tlie ends of these titbes containing the m ~ i p l e s\\-ere scratched a.ith a fili.. antl the tubes \\-ere then dried and ivciglietl. Tlie ends ivere then lirokcn off and tlie indiiuii trichloridc was tlisvlvcd in pure xvater. The solution \vas perfectlj- clcar in every case, hut it sometimes contained small particles of glass. This glass was reI1iO.i.i.d I)?. filtration and \\-as i\.eighcd, its weight being added to tliat of tlie cmpty tuhe aiid the broken off cntls of the tuhe. Thiel' states that filtratioii of the solution of the trichloride at this ponit was frequently necessar!. to reniovc ferric oxide tliat \vas present. During this investigatimi. !io\\ ever, tlie presence of ferric ositle in the sublimetl intliiun trichloritlr \v:is not observed in a!~!: case. (:ranis Illel,.

(:TalllS

Atolllic \I'eigilt

AgCI.

i,f Indinrn.

I

2.1156

4.11.42 I

1

7

4.9jy20

9,641 j6

rrj.8j

3 4

5 6

1.9sI 7-j

j.Sjj13 1.46561 4 .0S60z

,;, S j I 2 j I O . 77904

2.34557 7.91054

I?.SO

1'495 I L .